================
@@ -3107,6 +3107,23 @@ let TargetPrefix = "aarch64" in {
}
}
+ class SME_OuterProduct_TMOP_Intrinsic
+ : DefaultAttrsIntrinsic<[],
+ [llvm_i32_ty,
+ llvm_anyvector_ty,
+ LLVMMatchType<0>,
+ LLVMMatchType<0>,
+ llvm_nxv16i8_ty,
+ llvm_i32_ty],
+ [ImmArg<ArgIndex<0>>, ImmArg<ArgIndex<5>>,
+ IntrInaccessibleMemOnly]>;
+
+ def int_aarch64_sme_tmopa : SME_OuterProduct_TMOP_Intrinsic;
----------------
CarolineConcatto wrote:
I notice that we cannot make any difference between Za32 and Za16 when
lowering.
We need to have another aarch64_sme_tmopa. So maybe:
aarch64_sme_z32_tmopa
aarch64_sme_z16_tmopa
I would also change the other ones that are only needed for z32 bits to be:
s/ int_aarch64_sme_stmopa /int_aarch64_sme_za32_stmopa/g
s/ int_aarch64_sme_utmopa /int_aarch64_sme_za32_utmopa/g
s/ int_aarch64_sme_sutmopa /int_aarch64_sme_za32_sutmopa/g
s/ int_aarch64_sme_ustmopa /int_aarch64_sme_za32_ustmopa/g
https://github.com/llvm/llvm-project/pull/135145
_______________________________________________
cfe-commits mailing list
cfe-commits@lists.llvm.org
https://lists.llvm.org/cgi-bin/mailman/listinfo/cfe-commits
